Marco Valentino
- Postdoctoral Researcher, Idiap Research Institute
- marco.valentino@idiap.ch
Hello! I am a Postdoctoral Researcher at the Neuro-Symbolic AI Group at the Idiap Research Institute.
My main research activity lies at the intersection of Explainable AI and Natural Language Inference (NLI), investigating the construction of AI models that can reason in complex domains and generate natural language explanations in support of their decisions.
Specifically, I attempt to inject explicit domain knowledge into hybrid neuro-symbolic architectures to improve the generalization, robustness, and controllability of NLI models, and develop systematic evaluation methodologies to uncover reasoning strategies and limitations of large black-box neural networks.
Main Interests. Natural Language Explanations, Explanatory Reasoning, Neuro-Symbolic Models and Representations, Interpretability and Causal Analysis of NLP Models.
My main research activity lies at the intersection of Explainable AI and Natural Language Inference (NLI), investigating the construction of AI models that can reason in complex domains and generate natural language explanations in support of their decisions.
Specifically, I attempt to inject explicit domain knowledge into hybrid neuro-symbolic architectures to improve the generalization, robustness, and controllability of NLI models, and develop systematic evaluation methodologies to uncover reasoning strategies and limitations of large black-box neural networks.
Main Interests. Natural Language Explanations, Explanatory Reasoning, Neuro-Symbolic Models and Representations, Interpretability and Causal Analysis of NLP Models.
Publications
Here is a selected list of publications representative of my research work and Interests. For a complete list, check out my scholar profile!- Enhancing Ethical Explanations of Large Language Models through Iterative Symbolic Refinement, Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2024). Xin Quan, Marco Valentino, Louise A. Dennis, André Freitas
- Multi-Relational Hyperbolic Word Embeddings from Natural Language Definitions, Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2024). Marco Valentino, Danilo S. Carvalho, André Freitas
- Improving Semantic Control in Discrete Latent Spaces with Transformer Quantized Variational Autoencoders, Findings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2024). Yingji Zhang, Danilo S. Carvalho, Marco Valentino, Ian Pratt-Hartmann, André Freitas
- NLI4CT: Multi-Evidence Natural Language Inference for Clinical Trial Reports, Proceedings of The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023). Mael Jullien, Marco Valentino, Hannah Frost, Paul O'Regan, Donal Landers, André Freitas [website]
- Interventional Probing in High Dimensions: An NLI Case Study, Findings of the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2023). Julia Rozanova, Marco Valentino, Lucas Cordeiro, André Freitas
- Hybrid Autoregressive Inference for Scalable Multi-Hop Explanation Regeneration, 36th AAAI Conference on Artificial Intelligence (AAAI 2022, Oral). [code] Marco Valentino, Mokanarangan Thayaparan, Deborah Ferreira, André Freitas
- Case-Based Abductive Natural Language Inference, 29th International Conference on Computational Linguistics (COLING 2022). [code] Marco Valentino, Mokanarangan Thayaparan, André Freitas
- Diff-Explainer: Differentiable Convex Optimization for Explainable Multi-hop Inference, Transactions of the Association for Computational Linguistics (TACL, presented at EMNLP 2022). Mokanarangan Thayaparan, Marco Valentino, Deborah Ferreira, Julia Rozanova, André Freitas
- To be or not to be an Integer? Encoding Variables for Mathematical Text, Findings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022). Deborah Ferreira, Mokanarangan Thayaparan, Marco Valentino, Julia Rozanova, André Freitas
- Unification-based Reconstruction of Multi-hop Explanations for Science Questions, Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2021). [code] Marco Valentino, Mokanarangan Thayaparan, André Freitas
- Explainable Inference Over Grounding-Abstract Chains for Science Questions, Findings of the 59th Annual Meeting of the Association for Computational Linguistics (ACL 2021). [code] Mokanarangan Thayaparan, Marco Valentino, André Freitas
- Do Natural Language Explanations Represent Valid Logical Arguments? Verifying Entailment in Explainable NLI Gold Standards, Proceedings of the 14th International Conference on Computational Semantics (IWCS 2021). [data] [video] Marco Valentino, Ian Pratt-Hartmann, André Freitas
- A Framework for Evaluation of Machine Reading Comprehension Gold Standards, Proceedings of the 12th Language Resources and Evaluation Conference (LREC 2020), Marseille, France. [code] Viktor Schlegel, Marco Valentino, André Freitas, Goran Nenadic, Riza Batista-Navarro
Shared Tasks
- SemEval 2024 Task 2: Safe Biomedical Natural Language Inference, Accepted at the 18th International Workshop on Semantic Evaluation (at NAACL 2023). Mael Jullien, Marco Valentino, André Freitas [website] [competition] [starter kit]
- SemEval-2023 Task 7: Multi-Evidence Natural Language Inference for Clinical Trial Data, Proceedings of the 17th International Workshop on Semantic Evaluation. Best Task Paper Award Honorable Mention (SemEval at ACL 2023). Mael Jullien, Marco Valentino, Hannah Frost, Paul O'Regan, Donal Landers, André Freitas [website] [competition] [starter kit]
- TextGraphs 2022 Shared Task on Natural Language Premise Selection, TextGraphs-16: Graph-based Methods for Natural Language Processing (TextGraphs-16 at COLING 2022). [competition] [code] Marco Valentino, Deborah Ferreira, Mokanarangan Thayaparan, André Freitas, Dmitry Ustalov
- TextGraphs 2021 Shared Task on Multi-Hop Inference for Explanation Regeneration, Proceedings of the Fifteenth Workshop on Graph-Based Methods for Natural Language Processing (TextGraphs-15 at NAACL 2021). [competition] [code] Mokanarangan Thayaparan, Marco Valentino, Peter Jansen, Dmitry Ustalov
Workshops
- MathNLP: The 2nd Workshop on Mathematical Natural Language Processing, Accepted at The 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024). [website] Marco Valentino, Deborah Ferreira, André Freitas
- TextGraphs-17: Graph-based Methods for Natural Language Processing, Accepted at The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) Dmitry Ustalov, Arti Ramesh, Alexander Panchenko, Andrey Sakhovskiy, Elena Tutubalina, Gerald Penn, Marco Valentino
- MathNLP: The 1st Workshop on Mathematical Natural Language Processing, The 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP 2022). Deborah Ferreira, Marco Valentino, André Freitas, Sean Welleck, Moritz Schubotz [website]
- TextGraphs-16: Graph-based Methods for Natural Language Processing, 29th International Conference on Computational Linguistics (COLING 2022). Dmitry Ustalov, Yanjun Gao, Alexander Panchenko, Marco Valentino, Mokanarangan Thayaparan, Thien Huu Nguyen, Gerald Penn, Arti Ramesh, Abhik Jana [website]
Tutorials and Talks
- Reasoning with Natural Language Explanations, To be presented at The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024).
- Abstract Reasoning, Making Sense of ChatGPT. Idiap Research Institute.
- Transformers: Evoluzione e Limiti della Tecnologia Sottostante ChatGPT, Fra Antico e Futuro: La Filologia nell'Era dell'Intelligenza Artificiale. Sapienza University of Rome.
- Explainable Natural Language Inference, ITEE Seminars. University of Naples "Federico II".
- An Introduction to Deep Learning for Natural Language Processing, ITEE Seminars. University of Naples "Federico II".
- Hybrid Autoregressive Inference for Scalable Multi-hop Explanation Regeneration, The 36th AAAI Conference on Artificial Intelligence (AAAI 2022).